Temporal associations of ambient PM2.5 elemental concentrations with indoor and personal concentrations

Time series studies increasingly evaluate health relevance of the elemental composition of particles smaller than 2.5 mu m (PM2.5). Validation studies have documented that temporal variation of outdoor PM2.5 concentration is correlated with temporal variation of personal exposure, but very few papers have investigated the temporal correlation between outdoor concentration and personal exposure for the elemental composition of PM2.5. We evaluated the temporal association between outdoor concentration and personal exposure for the elements copper (Cu), zinc (Zn), iron (Fe), potassium (K), nickel (Ni), vanadium (V), silicon (Si) and sulfur (S) in three European cities. In Helsinki (Finland), Utrecht (the Netherlands) and Barcelona (Spain) five participants from urban background, five from suburban/rural background and five from busy street sites were selected (15 participants per city). Six outdoor, indoor and personal 96-h average PM2.5 concentrations were measured simultaneously in three different seasons (winter, summer and spring/autumn). Concurrently, samples were collected at a central reference site, reflecting urban background air pollution levels. The temporal variation at the central site was highly correlated with personal exposure for all elements, except Cu. The highest correlations (Pearson's R) were found for S and V (R between 0.87 and 0.98). Lower correlations were found for the elements Cu, Fe and Si associated with non-tailpipe traffic emissions and road dust (Pearson's R between 0.34 and 0.79). For PM2.5 mass the R was lower (between -0.37 and 0.70). Exclusion of observations most affected by indoor sources increased the personal to central site correlations but did not fully explain differences between elements. The generally high correlation between temporal variation of the outdoor concentration and personal exposure supports the use of a central site for assessing exposure of PM components in time series studies for most elements. The different correlations found for the eight elements suggests that epidemiological associations are affected by differences in measurement error. (C) 2014 Elsevier Ltd. All rights reserved.